Pre-1.0 · Current release v0.2.0

Where Ciaren is headed

This page describes the direction, not a dated release schedule. For exactly what's shipped, the code is authoritative and releases are tracked on GitHub.

View releases Read the docs

Project status: Pre-1.0, actively evolving

The public API, data model, and generated code may change between releases, with no backward-compatibility guarantee yet. Use Ciaren for experimentation, prototypes, and controlled internal workflows before relying on it for critical production jobs.

Open sourceAGPL-3.0Pre-1.0Plugin SDKLocal-first

Where it is today

The core platform works end to end.

Visual builder

66 transformation nodes plus file, SQL, and cloud-storage I/O — 80 built-in nodes in total.

Multi-engine execution

Polars (default) and pandas, selectable per run.

Data quality contracts

Assert not-null, unique, range, expression, and row-count checks.

Machine learning

Split, feature engineering, train, predict, and evaluate — with MLflow tracking and a Models page.

Scheduling

Built-in cron scheduler with retries, catch-up, and auto-disable.

Plugin platform

Versioned provider contracts, local and entry-point discovery, and signed .ciarenplugin packaging.

The direction

Themes, not dated promises

Ciaren is built to become a stable, local-first workflow platform with an extensible ecosystem around it.

Stable foundation

Pre-1.0 hardening — stabilize the public API, data model, generated code, and core node behavior before moving beyond 0.1.x.
Portable flow format — formalize .flow files with schema versioning, JSON Schema, and migrations.
Backend source of truth — move node metadata into the backend catalog, consumed via a complete catalog API.

Plugin ecosystem

Plugin lifecycle — keep improving discovery, loading, permissions, signature verification, install, update, and uninstall flows.
Community distribution — a lightweight index or marketplace for nodes, connectors, templates, exporters, and validators.
Production plugin examples — richer patterns beyond the hello-world node, including tests, packaging, and code export.

Connectors and data access

More connectors — expand database, file, API, and storage integrations while keeping credentials explicit and local-first.
Better browsing — improve remote file and table discovery for SQL, object storage, and local-folder connections.
Connection diagnostics — make test results, permission errors, and setup hints more actionable.

Exporters and portability

More export targets — explore notebooks, reusable job templates, and other portable artifacts.
Export validation — checks that generated artifacts run and match the visual flow's behavior.
Reusable handoff — make exported code easier to version, review, and run in another environment.

Data quality

Reusable contracts — strengthen validation nodes as first-class data contracts reusable across flows.
Quality reports — surface which checks passed or failed per run, with samples of problematic rows.
Validation exports — explore exporting quality checks into external test or validation formats.

Machine learning

ML workflow maturity — improve metrics, model comparison, lineage, and the Models page.
MLflow integration — clearer registered model workflows, aliases, and tracking configuration.
Guardrails — stronger warnings for data leakage, risky splits, and fragile training configurations.
ML templates — starter flows for common classification, regression, clustering, and feature-engineering tasks.

AI capabilities

AI as an extension point — introduced through providers, not as a required dependency of the core app.
Assistive workflows — pipeline generation, flow debugging, optimization suggestions, and plain-language error explanations.
Data-control safeguards — any AI integration stays explicit about what data is used and how users opt in or out.

Scheduling and automation

Schedule observability — better visibility into upcoming runs, missed runs, retries, and auto-disabled schedules.
Automation triggers — continue improving REST, CLI, and webhook-based ways to run flows outside the UI.
Lightweight orchestration — keep scheduling simple enough for local and self-hosted workflows.

User experience and documentation

Debuggability — better per-node errors, preview failures, and flow-level troubleshooting.
Onboarding — more demo projects, recipes, screenshots, and sample datasets for real workflows.
Contributor paths — make it easier to add nodes, connectors, exporters, validators, docs, and tests.

Staying focused

What Ciaren won't become in the short term

These are today's boundaries, not a lifetime promise — priorities may shift as the project and community grow, but they're not on the near-term roadmap.

A real-time streaming engine

Batch-style data and ML workflows are where the product is focused for now.

A warehouse-scale orchestrator

Heavy production orchestration currently belongs in dedicated tools — Ciaren exports clean code and integrates with them rather than replacing them.

A multi-tenant SaaS

The core project is local-first and self-hosted. A hosted layer, if it ever ships, would be opt-in and separate from the open-source core.

A black-box runtime

The core stays open-core: the engine and flow logic are inspectable, exportable, and yours to run. A specific plugin (say, a hosted AI feature) may wrap closed server-side logic — that's a plugin-level exception, not how the platform works by default.

Shape the roadmap

Ciaren is built in the open and the direction is decided in public — the best way to influence it is to get involved.

Discussions

Propose features and ideas.

Issues

Report bugs, request features.

Build a plugin

Extend the platform yourself.

Contribute

Good first issues are labelled.

Want to help shape this?

Star the repo, open a discussion, or ship a plugin.

View on GitHub Read the Docs

AGPL-3.0No account requiredRuns on your machine